Can Safeworld convince people that GenAI robots won’t hurt them?

Safeworld emerged from stealth with a seed round of more than $12 million, led by Shine Capital and a16z Speedrun, with additional investment from Box Group, Carnegie Mellon University Endowment, Innovation Endeavors and SV Angel.

The company was founded by Dr. Ding Zhao, who directs the Safe AI lab at Carnegie Mellon University, along with veteran startup executive Kyle Wong and machine learning engineer Simo Rachidi. Its stated aim is to evaluate robotic control systems in simulations populated with realistic human models.

Zhao described the safety challenge as a combination of advanced generative AI probabilistic evaluations and trust, saying both are needed to deploy a robot. Safeworld's approach involves building a digital version of environments such as a factory blind corner in a model like Genesis or MuJoCo, inserting a simulation of the robot being evaluated driven by its real software, and running thousands of scenarios in which human models encounter the robot.

Wong said tripping and falling is an example the company tests frequently in simulation, noting the alternative would be having someone trip and fall for the robot repeatedly. He also described questions such as what speed or stopping distance a robot needs to avoid colliding with a human, and whether a robot will detect a human carrying boxes.

Zhao argued this will be harder for robots than for vehicles because robots work in unstructured environments and each facility will have different safety standards. The founders believe robot-makers will want a third party to validate their work, including to share information about safety cases between competitors.

Vishal Dugar, CTO of Gritt Robotics, said his company is partnering with Safeworld as it develops safety simulations. He said most of its systems cannot be formally proven safe through math and equations, and verification necessarily has to be done empirically.

Safeworld is still figuring out the best model for its product, a platform for external users or a services-based approach.